Method and apparatus for resizing images using discrete cosine transform

• US 7,876,976 B2
• Filed: 02/23/2007
• Issued: 01/25/2011
• Est. Priority Date: 02/23/2006
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd., Seoul National University Industry Foundation (Seoul National University)
• Original Assignee: Samsung Electronics Co. Ltd., Seoul National University Industry Foundation (Seoul National University)
• Inventors: Cho, Nam-ik, Park, Chan-yul
• Primary Examiner(s): Mehta; Bhavesh M
• Assistant Examiner(s): Cunningham; Gregory F
• Application Number:
  US11/709,724
• Publication Number:
  US 20070237414A1
• Time in Patent Office: 1,432 Days
• Field of Search: 345/472, 345/472.1, 345/660, 345/FOR.149, 345/150, 345/220, 348/14.13, 348/445, 348/556, 348/581, 348/E5.055, 348/E5.111, 348/913, 358/1.2, 358/3.2, 358/451, 358/528, 358/FOR.143, 358/177, 375/E7.028, 375/E7.143, 375/E7.226, 375/E7.23, 382/166, 382/232, 382/233, 382/235, 382/244, 382/248, 382/250, 382/276, 382/298, 396/436, 702/88, 708/203, 708/208, 708/402, 715/252, 715/788, 715/798, 715/800, 715/815, 715/FOR.192
• US Class Current: 382/276
• CPC Class Codes: G06T 3/4084 in the transform domain, e.g. fast Fourier transform [FFT] domain scaling
• CPC Class Codes: H04N 19/117 Filters, e.g. for pre-processing or post-processing sub-band filter banks H04N19/635
• CPC Class Codes: H04N 19/18 the unit being a set of transform coefficients
• CPC Class Codes: H04N 19/48 using compressed domain processing techniques other than decoding, e.g. modification of transform coefficients, variable length coding [VLC] data or run-length data motion estimation in a transform domain H04N19/547; processing of decoded motion vectors H04N19/513
• CPC Class Codes: H04N 19/59 involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution
• CPC Class Codes: H04N 19/60 using transform coding

Abstract

A method for resizing an image using a resizing ratio may include receiving DCT (Discrete Cosine Transform) coefficients of an input image; calculating a transformation matrix for transforming the input image to an integer aspect ratio closest to the resizing ratio; performing a coarse resizing on the input image in a DCT domain using the transformation matrix; obtaining a spatial image by performing an IDCT (Inverse Discrete Cosine Transform) on the coarse-resized domain; and forming an output image by performing a fine resizing on the spatial image in a spatial domain.

Claims

1. A method of resizing an image using a resizing ratio, comprising:

receiving a DCT (Discrete Cosine Transform) coefficient matrix Y×

X of an input image;

calculating a transformation matrix for transforming the DCT coefficient matrix Y×

X of the input image using an integer aspect ratio closest to the target resizing ratio;

performing a coarse resizing on the DCT coefficient matrix Y×

X of the input image in a DCT domain using the transformation matrix;

obtaining a spatial image by performing an IDCT (Inverse Discrete Cosine transform) on the coarse-resized image; and

forming an output image having a coefficient matrix y×

x by performing a fine resizing on the spatial image in a spatial domain, wherein

the calculating of the transformation matrix includes;

calculating N′

that satisfies Y;

y=N;

N′ and

calculating M′

that satisfies X;

x=M;

M′

, where N and M are the number of vertical direction coefficients and the number of horizontal direction coefficients in a coefficient matrix of an DCT filter;

selecting integers N″ and

M″

closest to the calculated decimal fractions N′ and

M′

; and

calculating an N″

×

N″

transformation matrix and an M″

×

M″

transformation matrix using N″ and

M″

,

Y denotes the number of pixels in a vertical direction of the input image, X demotes the number of pixels in a horizontal direction of the input image, y denotes the number of pixels in a vertical direction of the output image, an x denotes the number of pixel in a horizontal direction of the output image, and

the performing of the coarse resizing includes calculating an N″

×

M″

transformation matrix of the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix.

2. The method of claim 1, wherein performing the fine resizing includes obtaining the output image by performing one of a bilinear filter and a bicubic filtering on the N″

×

M″

transformation matrix.

3. The method of claim 1, wherein the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix are formed by extracting a portion of DCT coefficients of an N×

M transformation matrix, and

wherein in performing the coarse resizing, the input image is downscaled by performing a decimation on the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix.

4. The method of claim 1, wherein the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix are formed by padding zeroes in the DCT coefficients of an N×

M transformation matrix, and

wherein in performing the coarse resizing, the input image is upscaled by performing an interpolation on the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix.

5. The method of claim 1, wherein the DCT filter is an 8×

8 DCT filter.

6. The method of claim 1, wherein the N″

and the M″

are integers closest to the values of N′ and

the M′ and

greater than the values of N′ and

M′

.

7. The method of claim 1, wherein the horizontal and vertical resizing ratios are the same.

8. The method of claim 1, wherein the horizontal resizing ratio is different from the vertical resizing ratio.

9. An apparatus for resizing an input image using a resizing ratio, comprising:

a receiver for receiving a DCT (discrete cosine transform) coefficients matrix Y×

X of an input image;

a transformation matrix calculation unit for calculating a transformation matrix for transforming the DCT coefficient matrix Y×

X of the input image using an integer aspect ratio closest to the resizing ratio;

a coarse resizing unit for performing a coarse resizing on the DCT coefficient matrix Y×

X of the input image in a DCT domain using the transformation matrix;

an IDCT (inverse discrete cosine transform) unit for obtaining a spatial image by performing an IDCT on the coarse-resized image; and

a fine resizing unit for forming an output image having a coefficient matrix y×

x by performing a fine resizing on the spatial image in a spatial domain, wherein

the transformation matrix calculation unit calculates N′

that satisfies Y;

y=N;

N′ and

calculates M′

that satisfies X;

x=M;

M′

, wherein N and M are the number of vertical direction coefficients and the number of horizontal direction coefficients in a coefficient matrix of an DCT filter, selects integers N″ and

M″

closest to the calculated decimal fractions N′ and

M′

, and calculates an N″

×

N″

transformation matrix an M″

×

M″

transformation matrix using N″ and

M″

,

Y denotes the number of pixels in a vertical direction of the input image, X denotes the number of pixels in a horizontal direction of the input image, y denotes the number of pixels in a vertical direction of the output image, and x denotes the number of pixels in a horizontal direction of the output image, and

the coarse resizing unit calculates an N″

×

M″

transformation matrix of the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix.

10. The apparatus of claim 9, wherein the fine resizing unit obtains the output image by using one of a bilinear filter and a bicubic filter on the N″

×

M″

transformation matrix.

11. The apparatus of claim 9, wherein the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix are formed by extracting a portion of DCT coefficients of an N×

M transformation matrix, and

wherein the coarse resizing unit downscales the input image by performing a decimation on the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix.

12. The apparatus of claim 9, wherein the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix are formed by performing a zero padding on an N×

M transformation matrix, and

wherein the coarse resizing unit upscales the input image by performing an interpolation on the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix.

13. The apparatus of claim 9, wherein the DCT filter is an 8×

8 DCT filter.

14. The apparatus of claim 9, wherein the values of N″

and M″

are integers closest to the values of N′ and

M′ and

greater than the values of N′ and

M′

.

15. The apparatus of claim 9, wherein the horizontal and vertical resizing ratios are the same.

16. The apparatus of claim 9, wherein the horizontal resizing ratio is different from the vertical resizing ratio.

17. A method of resizing an image using a resizing ratio, comprising:

receiving a DCT (Discrete Cosine Transform) coefficient matrix Y×

X of an input image;

calculating a transformation matrix for transforming the DCT coefficient matrix Y×

X of the input image using an integer aspect ratio closest to the target resizing ratio;

performing a coarse resizing on the DCT coefficient matrix Y×

X of the input image in a DCT domain using the transformation matrix;

obtaining a spatial image by performing an IDCT (Inverse Discrete Cosine transform) on the coarse-resized image; and

forming an output image having a coefficient matrix y×

x by performing a fine resizing on the spatial image in a spatial domain, wherein

the calculating of the transformation matrix includes;

calculating N′

that satisfies Y;

y=N;

N′ and

calculating M′

that satisfies X;

x=M;

M′

, where N and M are the number of vertical direction coefficients and the number of horizontal direction coefficients in a coefficient matrix of an DCT filter;

selecting integers N″ and

M″

closest to the calculated decimal fractions N′ and

M′

; and

calculating an N″

×

N″

transformation matrix and an M″

×

M″

transformation matrix using N″ and

M″

,

Y denotes the number of pixels in a vertical direction of the input image, X demotes the number of pixels in a horizontal direction of the input image, y denotes the number of pixels in a vertical direction of the output image, an x denotes the number of pixel in a horizontal direction of the output image.

18. The method of claim 17, wherein,

the performing of the coarse resizing includes calculating an N″

×

M″

transformation matrix of the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix

the performing the fine resizing includes obtaining the output image by performing one of a bilinear filter and a bicubic filtering on the N″

×

M″

transformation matrix.

19. The method of claim 18, wherein the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix are formed by extracting a portion of DCT coefficients of an N×

M transformation matrix, and

wherein in performing the coarse resizing, the input image is downscaled by performing a decimation on the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix.

20. The method of claim 18, wherein the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix are formed by padding zeroes in the DCT coefficients of an N×

M transformation matrix, and

wherein in performing the coarse resizing, the input image is upscaled by performing an interpolation on the input image using the N″

×

N″

transformation matrix and the M″

×

M″

transformation matrix.